AI GlossaryㅇWords you meet while using AI
AgentCore
An AWS infrastructure-as-a-service offering for running and managing AI agents in the cloud
In plain words
AgentCore is a service from AWS (Amazon Web Services) that runs and manages AI agents on the cloud. Here, an "agent" isn't just a program that answers questions—it's one that calls tools and handles multi-step tasks on its own.
Think of AgentCore as a "stage built just for agents." When an actor (the agent) needs more than just following a script—when it needs to grab props on the fly (external tools or local files)—AgentCore sets up the stage, turns on the lights, and makes sure the props end up exactly in the actor's hands, all behind the scenes. Developers only need to build the agent's "brain" (the logic that decides what task to do), and AgentCore takes care of actually running that reliably.
More recently, for agents running on AgentCore that need to access files or programs on a user's personal computer, a secure bridge structure connecting the cloud and the local machine has also come into use alongside it.
How it shows up in the news
In articles, you'll see phrasing like "the AgentCore Runtime hosts the Strands agent in the cloud and acts as an MCP client, sending tool discovery and invocation requests." One easy mistake is thinking AgentCore is the name of a specific AI model. It isn't—AgentCore refers to the cloud infrastructure that actually runs and manages agents, the stage itself, not a model.
See also
Stories using this term
- NVIDIA Swaps Harness, Lifts AI Agent Score from 30% to 100%AI · 2026.08.22
- AWS unveils 'Runtime Instances,' always-on compute for agentsAI · 2026.08.10
- LendingTree Builds Multi-Agent Mortgage Assistant on Amazon BedrockAI · 2026.08.09
- n8n Officially Supports AWS Bedrock AgentCore HarnessAI · 2026.08.09
- AWS unveils temporal policies to verify agent action historyAI · 2026.08.09
- AWS adds AI traffic rate limiting to AgentCore gatewayAI · 2026.08.09
